AI in Education: Evaluating the Efficacy and Fairness of Automated Grading Systems

Prof. Gulafsha Anjum, Prof. Jaya Choubey, Shubhanshu Kushwaha, Vandana Patkar · International Journal of Innovative Research in Science Engineering and Technology · 2023

The integration of Artificial Intelligence (AI) in educational settings has garnered significant attention, particularly in the realm of automated grading and feedback. Traditional grading methods are labor-intensive, timeconsuming, and prone to human bias, highlighting the need for AI-driven solutions to enhance grading efficiency, accuracy, and consistency. This study explores the application of machine learning algorithms and natural language processing techniques in developing automated grading systems. These systems can evaluate and score a wide range of student work, from multiple-choice assessments to complex written assignments, by identifying patterns and performance criteria from large datasets. The proposed method demonstrates a high accuracy rate of 97.6%, with a mean absolute error (MAE) of 0.403 and a root mean square error (RMSE) of 0.203. These metrics indicate the potential of AI to deliver quick, consistent feedback, allowing educators to allocate more time to personalized instruction and mentorship while making the educational process more scalable. However, challenges such as ensuring the transparency and fairness of AI algorithms, the need for substantial amounts of high-quality training data, and potential resistance from educators and students must be addressed. Additionally, AI systems must provide constructive and meaningful feedback beyond mere numerical scores to foster a supportive and effective learning environment. This paper reviews existing AI-based grading systems, assesses their efficacy and limitations, and discusses the ethical and practical considerations of their use. Through this comprehensive analysis, insights into how AI can enhance educational outcomes and transform traditional assessment and feedback paradigms are provided.

Read the paper · More papers on PaperTik